The noted secular trend toward an earlier age of pubertal onset in the developed world has been a subject of study for some time. The importance of energy balance and over- or under-nutrition is clear; a minimum level of energy availability is needed for puberty to ensue, but in contrast higher BMI is associated with earlier puberty (21). This latter statement has been seen especially in females (22, 23), but the underlying patho-biology is still not entirely clear. Leptin, a key metabolic hormone and modulator of BMI in humans, is produced from white adipose tissue (Figure 3). It is a major signal of energy sufficiency and mediates, at least in part, the influence of fat mass on pubertal timing. Leptin is a permissive signal for puberty and is necessary for normal reproduction. In females, serum leptin concentrations rise at the onset of puberty (26). Both humans and mice which lack leptin (Lep ob/ob) or its receptor (LepR db/db) show failure to complete puberty and are infertile (27). However, leptin is not the key coordinator in the up-regulation of GnRH signaling pathways at pubertal onset. Leptin alone does not stimulate pubertal onset and, whilst in females leptin concentrations rise during puberty, levels are lower in males and decrease during puberty (28). GnRH neurons do not express LepR therefore leptin cannot act directly to regulate GnRH neurons. Instead its acts indirectly via leptin-sensitive afferents which project to GnRH neurons (29). These afferents are likely to include LEPR-expressing GABA neurons from the arcuate nucleus, nitric oxide (which is required for its action) pathways, mTOR signaling, as well as kisspeptin/neuropeptide Y neurons (30, 31).

A key strategy in the attempt to uncover the key genetic regulators of pubertal timing in the general population has been genome-wide association studies (GWAS) of age at menarche and voice-breaking in healthy women and men, respectively. The first locus to be identified as associated with pubertal timing was the single nucleotide polymorphism (SNP) rs314276 in the gene LIN28B (32). The major allele of this SNP correlates with earlier breast development and menarche in girls (32). LIN28B is a human ortholog of a Caenorhabditis elegans gene important for developmental timing. The lin-28 family regulates, and is regulated by, the let-7 family of microRNAs (miRNAs). However, no human mutations in LIN28B have been identified, neither with delayed (33) nor with early puberty (34).

Since this initial discovery several increasingly large meta-analyses have been carried out on GWAS of timing of puberty. Whilst the first of these identified 42 (30 new, 2 previously confirmed and 10 possible) loci for age at menarche (35), an analysis of 182,416 European women encompassing 57 studies (36) identified 106 genomic loci. These meta-analyses are ongoing, but the largest to date which comprises 1000 Genomes Project-imputed genotype data in ~370,000 women has isolated 389 independent signals (P < 5 × 10⁻⁸) for age at menarche (37). The effect size of each of these alleles on the timing of menarche is between 1 week and 5 months. In total the loci identified in this study can explain ~7.4% of the variation in the timing of menarche in the general population, which corresponds to ~25% of the estimated heritability. Together this data suggests that individually many of these genetic variants have a low impact in the general population (37). Hence these huge studies suggest that there is a large degree of heterogeneity in the genetic determinants of normal pubertal timing. A large number of these signals show a significant association with Tanner staging in men and women, implying that the data is applicable to both genders. Additionally, many of these signals have been shown to have concordant effects on the age at voice breaking. However, in women the signals identified have stronger effects on early than on late age of menarche, but in contrast have larger effect estimates for relatively late than relatively early voice breaking in males (37).

Multiple signals in or near genes regulating the HPG axis function have been found by these studies including LEPR-LEPROT, GNRH1, and TACR3, mutations in which have been shown to be causal in CHH (38, 39). Loci in or near several further genes related to development of the pituitary and its function were also seen, including POU1F1, TENM2, and LGR4, the last of which acts as an enhancer for the pituitary development factor SOX2.

In addition to leptin signaling, several other genes implicated in body mass index including FTO, SEC16B, TMEM18, and NEGR1 have been implicated by GWAS as having a role in the timing of puberty. FTO had already been identified via GWAS of susceptibility to obesity, and it remains the original and most impactful locus with respect to effect on BMI and risk of obesity (40). Subsequently, using next generation sequencing techniques rare heterozygous variants in FTO have been identified in pedigrees with self-limited delayed puberty associated with extreme low BMI and maturational delay in growth in early childhood (41). In a parallel murine experiment, mice that were heterozygous for FTO gene knockout were shown to have significantly delayed timing of puberty (41).

A further gene, IRX3, identified at the same GWAS locus as FTO, was later found also to be of importance in influencing BMI (42); however the evidence from animal models on the effect of FTO on food intake regulation remains robust (43), although its actions may be complex (44). FTO-knockout mice (45) and in vitro studies have demonstrated that essential amino acids act to modulate the expression of FTO and that FTO acts downstream to influence mTORC1 signaling (46). mTOR acts as a coupler of energy balance and the activity of the reproductive axis by regulation of the hypothalamic expression of the kisspeptin gene (47). Blockade of mTOR in a rodent model led to delayed vaginal opening with blunting of the positive effects of leptin on puberty onset in food-restricted females (48). However, it is still unknown if the effect of FTO on pubertal timing is facilitated via effects on BMI, via mTOR signaling, or by both.

Neuropeptide Y (NPY) is another protein implicated in the regulation of food intake and satiety, as well as the hypothalamic-pituitary axis. NPY increases the response of pituitary gonadotrope cells to GnRH (49), both by stimulating GnRH binding to pituitary GnRH receptors and by its action upstream at the median eminence to potentiate GnRH secretion from GnRH axon terminals (50). Studies with primate models imply that NPY may contribute to the brake that restrains the onset of puberty between infancy and mid-childhood (51). The link between energy homeostasis and reproductive development may also be mediated by ghrelin and other gut-derived peptides (52–54). α-MSH signaling via MC3/4 receptors, acting to increase Kiss1 expression and mediate the permissive effects of leptin on puberty, has also been implicated recently as an important element in the metabolic control of puberty (55). Lastly, mice lacking the insulin receptor in astrocytes have delayed puberty and irregular estrus cycles, with reduced astrocyte prostaglandin E synthase 2 levels (56). However, roles for the majority of these genes involved in fat mass and metabolic regulation have not been demonstrably shown in human delay of puberty. A small cohort of 31 patients was analyzed for mutations in the ghrelin receptor, or GHSR, and 5 patients were found to have point mutations in this gene (57).

It is biologically very plausible that the pathophysiology of delayed puberty and conditions of GnRH and gonadotropin deficiency share a common genetic basis. Therefore, investigations have been carried out into the role of genes known to cause CHH in the phenotype of isolated delayed puberty. Previous studies in CHH cohorts have found mutations in HS6ST1, FGFR1, and recently in KLB, not only in small numbers of patients with CHH but also in their relatives with delayed puberty (58–60). Last year, a study was completed that aimed to compare the frequency with which mutations in genes (n = 24) known to cause GnRH or gonadotropin deficiency were found in patients with CHH and individuals with self-limited delayed puberty. This comparison found a significantly higher proportion of mutations in the CHH group (51% of CHH probands vs. 7% of delayed puberty probands, p = 7.6 × 10⁻¹¹). Whilst this is perhaps unsurprising, a greater degree of oligogenicity in these GnRH deficiency genes was also seen in the CHH group, suggesting a mostly distinct or as yet undiscovered genetic basis of these two conditions (61). Mutations in Kallmann Syndrome (KS) genes such as ANOS1 and NSMF, leading to hypogonadotropic hypogonadism with anosmia, have not been found in individuals with self-limited pubertal delay.

Studies using next generation sequencing to examine cohorts of patients with delayed puberty have identified variants in several CHH genes, particularly GNRHR, TAC3 and its receptor TACR3, but also in IL17RD and SEMA3A (62). However, these variants have not been tested in vitro or in vivo for pathogenicity, or investigated for within pedigree segregation. Many syndromic conditions have delayed or absent puberty within the phenotypic spectrum of the condition, see Table 1.

Recently, using whole and targeted exome analysis a mutation in HS6ST1 was found in one extended pedigree from a large cohort of patients with isolated familial delayed puberty, for the first time without associated CHH in patient relatives (72). All of the six family members in three generations that carried the mutation had a classical self-limited delayed puberty phenotype, with no individuals displaying CHH. A spontaneous onset of puberty was seen in the proband at 14.3 years. A mouse heterozygous knockout model was also examined in parallel. This work substantiated that loss of one allele of Hs6st1 can provoke pubertal delay but with normal adult reproductive capacity. The Hs6st1^+/− mice displayed no compromise in their fertility, GnRH neuron or testes development or spermatogenesis and were born at normal Mendelian ratios. However, female mice were seen to have a significant delay in the timing of vaginal opening, a surrogate for onset of puberty in female rodents.

Notably the Hs6st1^+/− mice had no defects of olfactory bulb morphology and no significant reduction in the total number of GnRH neurons in the hypothalamus or extending to the median eminence to explain the pubertal delay. Instead, this might be mediated by changes in either GnRH neuron activity or other relevant downstream pathways, implied by the expression of Hs6st1 mRNA in both the arcuate nucleus and paraventricular nucleus (73, 74). These results indicate whilst, as above, many patients with familial self-limited delayed puberty do not carry mutations in CHH genes, perturbations in a single allele of a particular subset of genes that modulate the HPG axis may be enough to result in a phenotype of self-limited pubertal delay. In contrast, more deleterious alterations in these genes, mutations in both alleles of a gene or a heterozygous mutation in combination with mutations in further genes, are needed to produce the more severe phenotypes of CHH and KS (75).

A further study utilizing whole and targeted exome sequencing methods in the same large Finnish cohort of individuals with familial self-limited delayed puberty, identified deleterious mutations in the IGSF10 gene in six unrelated families (76). Mutations in this gene affect the migration of GnRH neurons from the vomeronasal organ in the nose to the forebrain during embryonic development (Figure 4). The patients with these mutations presented in adolescence with pubertal delay without features of constitutional delay in growth. Given that a functional GnRH neurosecretory network is required for the onset of puberty, the hypothesis produced from this work is that disruption of GnRH neuronal migration, as caused by aberrant IGSF10 signaling, could result in arrival of fewer (or delayed) GnRH neurons at the hypothalamus. This would then in turn lead to a functional defect in the GnRH neuroendocrine network and an increased “threshold” for the onset of puberty, with a resultant delay. In addition, loss-of-function mutations in IGSF10 were found in patients with a hypothalamic amenorrhea-like phenotype, implying a shared genetic basis of functional central hypogonadism with both CHH (77) and delayed puberty. However, although deleterious mutations were enriched in CHH patients, there was lack of complete segregation with trait in these permanent hypogonadotropic hypogonadism families, suggesting that haploinsufficiency of IGSF10 is not sufficient to cause this phenotype. Interesting, mutations in IGSF10 have very recently also been found in patients with both premature ovarian insufficiency and disorders of neuronal development, and in the same report in a further pedigree with a Kallmann-like phenotype (78). The results of the studies on HS6ST1 and IGSF10 in delayed puberty point to a mechanism by which developmental defects in the GnRH system during fetal life can modulate the timing of pubertal onset in adolescence, seemingly without other phenotypic features. It remains to be determined whether these patients have any deficiency of the their long-term reproductive capacity or sexual lifespan.

Autosomal recessive CHH is most frequently caused by loss-of-function mutations within the GnRH receptor, accounting for 16–40% of this patient group. Mutations have been found within the extracellular, transmembrane, and intracellular domains of the receptor leading to impaired GnRH action (79). Sequencing studies that have analyzed the GNRHR gene in cohorts with self-limited delayed puberty (79), have found just a handful of deleterious mutations. A homozygous partial loss-of-function mutation in GNRHR was found in two brothers, one with self-limited delayed puberty and one with CHH (80), and a further heterozygous mutation found in one male with self-limited delayed puberty (81). Far more rarely defects of the glycoprotein hormones luteinizing hormone (LH) or follicle-stimulating hormone (FSH) can lead to CHH, in particular via mutations in the specific β-subunits (82, 83). In women, loss-of-function mutations of LHβ result in normal or late timing of menarche (following a normally timed onset of puberty) but with later infertility resulting from lack of ovulation (84). In men, similar defects lead to a presentation with absent pubertal development secondary to Leydig cell hypoplasia resulting in testosterone deficiency and failure of spermatogenesis. Women with inactivating FSHβ mutations display pubertal arrest and primary amenorrhea whilst men have a similar pattern of spontaneous entry into puberty followed by arrest with azoospermia (85). Defects in these two genes do not usually present with a classical picture of self-limited delayed puberty.

Overall, from the evidence we have from current published work we can conclude that although there are some shared gene defects, the genetic basis of CHH and delayed puberty is likely to be due to different, currently unrecognized, genes in many cases (Table 2) (81).

However, there is a wide spectrum of phenotypes in patients with central hypogonadism, ranging from complete hypogonadotropic hypogonadism, with failure of pubertal development, to partial hypogonadism with an arrest of pubertal development, and even reversible hypogonadotropic hypogonadism in some patients post treatment (87–90). It may be a prudent strategy for clinicians to focus the use of genetic testing of known CHH genes in delayed puberty patients on those patients with either extreme delayed puberty, clear familial inheritance or red flags (such as micropenis, cryptorchidism, anosmia, cleft lip or palate or renal agenesis) which would point to a syndromic or CHH phenotype.

The central control of pubertal onset, after the mid-childhood period of HPG axis quiescence, is orchestrated by a resurgence of the GnRH pulse generator, with a reduction in central inhibition and a sharp upregulation in the activity of this axis. This activity is permitted by a change in the balance of GABA-glutamate signaling in the brain (91). Around this time morphological changes in GnRH neurons have been observed including increases in dendritic spine density and a simplification of their dendritic architecture. The intensification of kisspeptin signaling in the hypothalamus, one of the key hormonal players in puberty onset, at this time is a consequence of both an increase of kisspeptin synthesis and a rise in the responsiveness of GnRH neurons to kisspeptin stimulation. This mechanism has been well observed in murine models, but also in primates and is relatively conserved during evolution (92). However, what is far less well understood are what the triggers are for this upregulation of kisspeptin biosynthesis in the hypothalamus at the end of the juvenile period. Thus, whilst there is strong evidence that the secretion of kisspeptins from KNDy neurons in the arcuate nucleus is one of the vital stimulatory inputs on the GnRH pulse generator, it is not likely to be the ultimate controller of the release of the puberty brake. Rather, kisspeptin is the conductor of the orchestra of upstream stimulators and repressors influencing the system at this crucial developmental stage (51).

It is likely, therefore, that there is no one single gene that is capable of the hypothalamic control of puberty onset. Instead, we can imagine a hierarchical network of genes acting together to lift the brake applied during the dormancy of the HPG axis in mid-childhood (Figure 3). Data to support this hypothesis have largely come from a systems biology approach (93) and animal models (51), with little data from human subjects. It is clear that transcriptional repression is fundamentally important to the regulation of gene expression in mammals. Transcriptional repressors containing zinc finger motifs, which recognize specific DNA sequences in regulatory regions of the genome, are particularly appealing candidates to have major roles in this governing network (94). Potential key regulators include Oct-2, Ttf-1, Yy1, and Eap1. Oct-2 is a transcriptional regulator of the POU-domain family of homeobox-containing genes. Oct-2 mRNA is upregulated in the hypothalamus in juvenile rodents; blockage of Oct-2 synthesis delays age at first ovulation and hypothalamic lesions which induce precocious puberty (e.g., hamartomas) activate Oct-2 expression (95). Ttf-1 is another homeobox gene that enhances GnRH expression (96). Ttf-1 expression is increased in pubertal rhesus monkeys (97). Yy1 is a zinc-finger transcription factor with crucial roles in normal development and malignancy (98). Eap1, or Enhanced at puberty 1, codes for a nuclear transcription factor, characterized by a dual transcriptional activity: it both trans-activates the GnRH promoter, which facilitates GnRH secretion, and inhibits the preproenkephalin promoter, which represses GnRH secretion. Eap1 mRNA levels increase in the hypothalamus of primates and rodents during puberty, and Eap1 knockdown with siRNA causes delayed puberty and disrupted estrous cyclicity in a rodent model (99–103). Therefore, Eap1 transcriptional activity facilitates the initiation of female puberty, in a manner that is independent of hypothalamic Kiss1 expression (101). Eap1 gene expression is itself regulated by both activation by Ttf-1, and repression by Yy1 and a further transcriptional regulator Cux1 (104).

A very recent discovery is of the first human EAP1 mutations that appear to be causal for self-limited delayed puberty in two families (86). The affected individuals from these two families had classical clinical and biochemical features of self-limited delayed puberty, with presentation at more than 15.5 years with delayed onset of Tanner stage 2 and delayed peak height velocity. Both probands had spontaneous pubertal development by the age of 18 years without testosterone therapy, thus excluding CHH. By whole exome sequencing of probands with familial delayed puberty two highly conserved variants—one in-frame deletion and one rare missense variant in EAP1—were identified. Using a luciferase reporter assay, EAP1 mutants showed a reduced ability to trans-activate the GnRH promoter compared to wild-type EAP1, due to reduced protein levels caused by the in-frame deletion and sub-cellular mis-location caused by the missense mutation. This study also demonstrated by chromatin immunoprecipitation that EAP1 binding to the GnRH1 promoter increases in monkey hypothalamus at the onset of puberty.

Furthermore, evidence from a recent study has emphasized the importance of the transcriptional control of the Kisspeptin gene Kiss1. This regulation by the polycomb complex proteins EED and Cbx7 is thought to be an important transcriptional repressive mechanism to prevent the premature onset of puberty (105). In the latter stages of mid-childhood there is an increase in the methylation of the promoters of these genes, resulting in a reduction in expression, as well as a decrease in the binding of EED on the Kiss1 promoter. This inhibition of Kiss1 repression also correlates with reduced expression of transcription factors containing certain zinc finger motifs. Moreover, there is also reorganization of the chromatin status and changes in histone methylation to accompany the loss of these polycomb complex proteins from the Kiss1 promoter (106). Studies on both rats and goats also provide data on changes in histone acetylation and gene methylation resulting in alterations in gene expression during puberty (107, 108).

There are a number of different epigenetic mechanisms that may have importance for the regulation of the pubertal timing, including imprinting. Imprinted genes are known to influence the timing of several key developmental stages in humans including weaning and adrenarche. In general, paternally expressed genes promote later childhood maturation and maternally expressed genes promote a more premature maturation (109). This holds true for two paternally inherited genes, MKRN3 and DLK1, which are associated with age at menarche in girls and voice-breaking in boys from the GWAS discussed above (37). Variants in both of these genes have been discovered in patients with familial central precocious pubertal timing, with paternally-inherited mutations leading to the expression of the phenotype (110, 111). MKRN3 is thought to contribute to the puberty brake restraining the HPG axis via inhibition of GnRH release. This gene encodes Makorin Ring finger protein 3, a zinc finger protein containing a C3HC4 motif (known as a RING domain) associated with E3 ubiquitin ligase activity (112, 113). Since MKRN3 expression in the arcuate nucleus falls in murine models between birth and weaning, and in humans serum concentrations decline at puberty onset, it is thought to have an inhibitory effect on the GnRH network (114, 115). This supports the hypothesis that the onset of puberty is a consequence of the removal of gonadotropic axis repression. However, what is still unclear is where MKRN3 is placed in this hierarchy of gene regulators controlling kisspeptin levels. Very new data has demonstrated that knock-out of MKRN3 in pluripotent stem cells does not affect GNRH1-expression when these cells are later differentiated into neurons (116). In terms of delayed puberty, mutations in neither MKRN3 nor DLK1 genes have been described in human patients with these conditions.

Prader-Willi syndrome (PWS) is frequently caused by disorders of imprinting and is often associated with either absent or delayed puberty (117). In most patients with PWS the syndrome is due to a deletion of a cluster of imprinted genes (including MKRN3) on the paternally inherited copy of chromosome 15 (paternal deletion), or by inheritance of both copies of this cluster from the mother (maternal uniparental disomy) (118). Precocious puberty is relatively uncommon in PWS (119), but most individuals show some degree of pubertal failure, with one or a combination of an absent pubertal growth spurt, hypogonadotropic hypogonadism, cryptorchidism, underdeveloped genitalia, or primary amenorrhea (120). The probable explanation for the rarity of precocious puberty in individuals with PWS, despite the lack of MKRN3 expression, is the effects of other genes inactivated by the imprinting defect, in particular MAGEL2 (121, 122). This points to a complex role for imprinted genes in the pubertal timing, with tissue type and developmental stage specific gene expression (109, 118).

Evidence from murine models has demonstrated that non-coding RNAs can act as epigenetic modulators of the timing of puberty. Specific microRNAs play a role in the epigenetic up-regulation of GnRH transcription during what is known in mice as “the critical period,” or infantile mini-puberty in humans (123). A key pair of microRNAs (miR-200 and mIR-155) are thought to regulate Gnrh1 expression, and to control the expression of two important transcriptional repressors of Gnrh1, Zeb-1, and Cebpb. There is an associated increase in the transcriptional activation of GnRH1 with a reduction in Zeb-1 and Cebpb, the latter a nitric oxide-mediated repressor of Gnrh1 that acts both directly and through Zeb1. These changes lead to the up-regulation of Gnrh1 synthesis in GnRH neurons (123). Moreover, miR-7a2 has been demonstrated to be essential for normal murine pituitary development and HPG function, with deletion in mice leading to hypogonadotropic infertility (124).

The increase of kisspeptin and GnRH expression in the hypothalamus at puberty is therefore the result of the actions of an intricate arrangement of repressing and activating transcription factors controlling Kiss1 and GnRH1 transcription, with these being themselves influenced by several epigenetic mechanisms including DNA methylation, histone modification and non-coding RNAs (106, 123, 125, 126). Moreover, these epigenetic mechanisms are possible facilitators of gene-environment interactions that also have influence on the hypothalamic regulation of puberty. A number of different sources of evidence have demonstrated that the brain epigenome at puberty is affected by environmental disturbances (107). Endocrine-disrupting chemicals (EDCs), often found in products commonly used in the developed world, have been considered as a potential cause for pubertal timing disturbance for many years, with increasing concern among the lay population (127). Many and varied substances have been identified as possible EDCs, such as polybrominated biphenyls, bisphenol A, atrazine (herbicides), and glyphosate, but also common medicines including paracetamol and betamethasone (128–131). It has been observed that adolescents who have been exposed to the estrogenic insecticide DTT and then adopted internationally display early or precocious pubertal timing (132).

The most important timing of EDC exposure for impact on pubertal timing was historically considered to be in late childhood, but there is now clear data that there may be prenatal and infantile origin of alterations in the timing of puberty. In utero exposure in males to EDCs, in particular to phthalates, can result in under-masculinization of genitalia (133). Moreover, exposure of pregnant rodents to EDCs has been associated with epigenetic alterations in testis as well as other systemic effects. This together suggests that epigenetic changes in the fetal period are a potential mechanism for the hypothalamic effects of prenatal exposure to EDCs (131). These effects may manifest in pregnant rodents, their unborn fetus but also into the next two or more generations as well (134).

However, it is difficult to definitively demonstrate a mechanism of action for EDCs through the premature activation of the hypothalamic GnRH pulse generator. Recently, exposure of female mice to arsenic in utero was shown to alter the hypothalamic expression not only of GnRH and LH but also of their upstream transcriptional regulators, in particular Oct-2 and Ttf-1 (135). Mice exposed to arsenic demonstrated precocious puberty with premature vaginal opening, a marker of the onset of puberty rodents. However, in most datasets it has been difficult to unpick the most likely differing, and possibly conflicting, influence of varying doses and combination of EDCs affecting estrogenic, androgenic or other pathways, and changes in effects depending on age and length of exposure (136).

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.